Subject: Re: Strange system hang (amd64)
To: Paul Goyette <email@example.com>
From: Quentin Garnier <firstname.lastname@example.org>
Date: 05/27/2007 20:57:24
Content-Type: text/plain; charset=us-ascii
On Sun, May 27, 2007 at 11:46:56AM -0700, Paul Goyette wrote:
> I've got this very wierd situation going on...
> One of my amd64 (zippy) boxes mounts /usr/obj via NFS from the other=20
> box (quicky). Both are currently running 4.99.18, so it's time to=20
> update to more -current.
> So, I log in on zippy, and attempt to copy the newest kernel file from
> the NFS-mounted /usr/obj/... filesystem to my root filesystem. It=20
> promptly creates the output file but then immediately hangs hard. So=20
> hard that I can't even do an 'ls -l /' command from another session!
> Attempting to figure out where the problem is, I've reboot both the NFS=
> server as well as the client. No improvement - hangs in the same place.
> I've done the following tests
> cp /2MB-file /2MB-copy Succeeds
> cp /2MB-file /dev/null Succeeds
> cp /usr/obj/2KB-file /2KB-copy Succeeds
> cp /usr/obj/2KB-file /dev/null Succeeds
> cp /usr/obj/2MB-file /dev/null Succeeds
> cp /usr/obj/2MB-file /2MB-copy Fails
> I haven't yet tried copy in the local-FS --> NFS direction yet, but so=20
> far this problem seems to affect only copying of large files from the=20
> NFS file system to the local fs. (I'm currently running memtestplus=20
> just to eliminate the possibility of memory failure.)
> Any clues on what's wrong? Or what else I can do to help narrow this=20
tcpdump and see what goes on. It might be as simple as a MTU issue, or
indeed a kernel bug. You can play with the mount options anyway,
reducing the nfs packet size, switching to TCP, that kind of stuff.
Quentin Garnier - email@example.com - cube@NetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2, 2005.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (NetBSD)
-----END PGP SIGNATURE-----